A Hot Segmentation Tool: The Channel Slicer

May 13, 2008 1:08 AM  By

There are thousands of ways to segment a house file. But one of the best is to do it by channel. How did your customers come in—and how did they order?

Here is a three-step process for slicing your data this way.

Step 1: Consolidate multichannel permutations into actionable segments.

One can get stuck in analysis paralysis if all possible permutations regarding channel are considered. Here’s the way it’s usually done:

a. Catalog (i.e., mail / phone / Fax), Internet and retail store
b. Catalog, Internet
c. Catalog, retail store
d. Internet, retail store
e. Catalog only
f. Internet only
g. Retail store only

Ugh, What a nightmare. Can you imagine how many unique segments that would create when combined with RFM variants?

Below is a better approach:

Catalog: the combination of all customers falling into groups a, b, c, and e.

Internet: the combination of all customers falling into groups d and f.

Retail store: all customers falling into group g.

This system allows multichannel retailers to isolate customers who have never placed orders via mail, phone or fax. Not surprisingly, it is precisely these groups of customers that produce limited incremental ROI when mailed against holdout panels.

Step 2: Determine some common sense exceptions to the rules.

Even the best of constitutional documents need an amendment or two. Below are a few to ponder:

Internet orders where the catalog quick shop feature was used should, instead, be flagged as catalog orders.

Internet orders where items with SKU prefixes found only on merchandise displayed in printed catalogs are ordered should, instead, be flagged as catalog orders.

Internet orders where a catalog-attributable key code is provided should, instead, be flagged as catalog orders.

Catalog buyers who have not used the channel in some reasonable timeframe—two or three years, let’s say—should be classified under a more recently-used channel, if applicable.

In the merge, prioritize retail store only buyers below inquiries and perhaps even below gift recipients/ship-to addresses. Our observation, to date, has been that catalog requesters produce a greater ROI on print advertising investment than retail store buyers, assuming equivalent recency. This pattern, though, has not been consistently observed with regards to giftees/ship-to’s.

Step 3: Develop a contact strategy that maximizes print advertising ROI.

The key measure here is incremental sales per catalog mailed, and here are some tests to consider:

Seasonally, reduce the number of contacts to Internet buyers with similar RFM attributes to their catalog buyer counterparts by one, or, longitudinally, reduce the number of contacts to Internet buyers by 30 – 50%.

Understandably, this is a fairly bold test I am suggesting—it can be executed, though, with minimal risk to your on-going operations. Just be sure your holdout panel quantities will produce statistically significant results.

In comparison to circulation plans that were executed in years past without channel segmentation, mail deeper into catalog buyers and, conversely, less deep into Internet buyers. All else being equal, it is a safe assumption that customers who have a proven track record of responding to catalog offers will perform better than those who never have.

Do not neglect your best-of-breed retail store only buyers. Time and time again, we see excellent incremental response from high frequency, high average order, repeat retail store only buyers.

If your marketing team is skeptical, or if your store group cries cannibalization, conduct a test where the holdout panel is the control and the mailed portion is the test cell—and further allay their fears by slapping a front cover dot whack driving business to the store in question.

Todd Miller is a director with San Rafael, CA-based catalog consultancy Lenser.